X
Business

Solving outages with application performance monitoring

Q&A with AppDynamics European CTO Firaas Rashid.
Written by Colin Barker, Contributor

AppDynamics helps companies keep their systems running with its application performance management (APM) software that monitors their applications and provides customers the information they need to fix problems quickly. The firm was acquired by Cisco for $3.7bn last year. ZDNet talked to Firaas Rashid, AppDynamics CTO for EMEA, to find out more.

ZDNet: Tell me about AppDynamics

Rashid: If you are familiar with modern day APM, you will understand how it works. You basically install it on any software and by installing it you don't change any code.

You literally start up a JVM with AppDynamics as a parameter. Or you start up AppDynamics on a .NET server and click on the different parameters you want to monitor and basically out-of-the-box it automatically discovers the whole application architecture, including every database you write to, message queue and so on, and it tracks every transaction that's going through. It baselines it, tells you when things are deviating from the norm and, when you've got a problem, it tells you the root cause down to the code level.

See: Digital transformation: A CXO's guide

Imagine in the banking environment. We [Rashid previously worked for a financial services company] often have problems and we hit a peak in the number of transactions that are taking place and things are starting to slow down. We wouldn't know about it until the traders were telling us. And then, when they did slow down, we wouldn't find the root cause for hours.

AppDynamics does all of this with no configuration, no code changes, and it does it in a very easy-to-use user interface as well.

So the reason that people grew very passionate about it from the client side was that we solved some very, very personal pain for people.

For me, I was woken up regularly at three o'clock in the morning and found myself jumping out of bed, trying to find a dark corner in my house to take the call and then the call could go on for, like, five hours.

In the meantime, you're trying to get in touch with developers, DBAs, Unix guys, trying to find out where the problem is.

Then it takes so long to find out the root cause and it turns out to be some little component that's been knocked over.

For me, the big thing was that it that AppDynamics very quickly began to take those five hours down to less than an hour and eventually to a few minutes.

They like it for its speed and efficiency?

Yes, but the other thing it does is it actually drives people's careers.

You see, not only do we help solve these massive outages in a very simple and meaningful way, for the first time when people start to show what they are doing using AppDynamics the business often says, that's the first time I've seen someone in IT talk about a business process in a very simple way for me to understand in terms of technology.

And we find that consistently when people bring in AppDynamics to drive the business they end up having their careers accelerated as a result.

And so that was the basic idea, that you could deliver real-time access to every aspect of a business?

And end-to-end. The traditional way of doing monitoring was always, monitor this CPU, the disc space and network, and the problem was that you were always solving different problems but not the main problem.

The way that [the founders] built the company was on the understanding that you've got to build everything from the perspective of the user. So from the mouse-click we'll start measuring how that follows through the different components and then into the database and then back out again and we'll put a time against that. So now, if that slows down, we know that there's a problem. It doesn't matter if the CPU spikes or the memory spikes or anything else happens. We need to understand what's happening with the business.

That's the way they sought to do it and they were also passionate about building enterprise software in a new way. And they were one of the first companies ever to have enterprise software downloadable on their website as a free trial. Why? Because it was so easy to get up and running with it, and you didn't need all of the customizations and modifications that people typically associate with enterprise software.

They had the confidence needed to do that?

Exactly and still today you can go online and download it in a trial. You don't have to interact with a human, you just load it and install it.

When we go into clients for the first time and do a proof of concept with them, they are often up and running within a few hours. They can have a rest environment up and running and shortly afterwards it will be in production.

And we do this across all sorts of industries. We have airlines like easyJet and United, the automotive industry - and a lot of them are running IoT clients so that they are monitoring the end-to-end performance of the connected car.

What are the other issues here?

One of the things we are focused on is business performance. As well as showing you the end-to-end transaction performance, we are also showing the users and the numbers that are flowing through.

Take the insurance market, where now we can show how many people are flowing through when looking for mortgages and how many people are dropping out. We now have a belief that every company is basically a software company and that if the software goes the entire business collapses.

There is a large bank that uses our dashboards to measure their end-to-end performance, including how many people are coming in, what type of customer is coming in and then they can see what is happening and ask questions like, "Are there any changes that we need to make right now to make our technology fit to drive the revenue we need to do?".

There is an ability to measure what is happening and to deal with issues right now, before they become a problem?

We have a concept called a baseline and that's a self-learning product that we've built which is based on the time of day, the time of week or the time of month.

So, we learn what "normal" looks like and that is normally based on any statistic - it could be revenue. We can then say that normally at this point you would make a £100 pound and you've made £80 and we can tell you that; but what that means is that we are not telling you when things are starting to break but when they are starting to look as though they might break.

Can you give me a practical example of how the technology is used?

Vodafone is probably the best one. The biggest day this year for a company like Vodafone is iPhone launch day.

Vodafone brought us in a month or so before the iPhone 7 launch and said, "This has got to work, there is no question that it can't work".

They had us deployed in their environment and what they were immediately able to see was where the bottlenecks were. They were able to see as people where traversing the system, where things were starting to slow down.

Was it at their check-out page? Or was it later on when they were trying to insert their account details?

We had Dixons Carphone who brought us in for the same reason. What they wanted was to know that their systems could cope with the demand before they even got to iPhone launch day.

Then they got us in again for Black Friday. They looked at the same issue, are we able to handle the transactions both at the store and online.

How do you see things developing in the future?

We are heavily focused on the business impact of technology, so we are trying to understand where people are trying to use technology and whether it works or whether it doesn't. Who are those people using the technology and what are they worth to the business? What are they trying to do?

Because we think we need to understand that both from the code level up to the business impact of that code. So that's part of our Business iQ platform.

And we see take-up of that almost everywhere whether it's banks driving mortgage journeys or retail companies looking at what type of customers they have and what their requirements are.

Another thing is we made an acquisition of a company called Perspica and what they are doing is in machine learning and artificial intelligence. So we are working that product into AppDynamics

That's because we want to be much more active in predicting the root cause when things go wrong.

More on managing system performance:

Google Cloud Platform adds duo of application performance management tools for developers

Google Cloud Platform is rounding out its developer tools for application performance management with Stackdriver Profiler, service that allows you to see how code actually performs in production.

Fortinet Security Fabric merges with IBM Threat Management system

At the RSA conference IBM and Fortinet announced that they have expanded their strategic relationship by bringing the capabilities of their security offerings together.

VMware updates vSphere, vSAN with more hybrid management features

With the latest releases of vSphere and vSAN, VMware is continuing to build up support for customers interested in hybrid cloud options.

Boston Venom EPYC Workstation review: Price/performance equation adds up for multi-threaded applications

This dual-socket AMD EPYC-powered system offers a total of 64 cores and 128 threads, giving it an advantage over comparable Xeon-based alternatives for many types of workload.

Editorial standards